Global Optimization with Sparse and Local Gaussian Process Models

We present a novel surrogate model-based global optimization framework allowing a large number of function evaluations. The method, called SpLEGO, is based on a multi-scale expected improvement (EI) framework relying on both sparse and local Gaussian process (GP) models. First, a bi-objective approach relying on a global sparse GP model is used to determine potential next sampling regions. Local GP models are then constructed within each selected region. The method subsequently employs the standard expected improvement criterion to deal with the exploration-exploitation trade-off within selected local models, leading to a decision on where to perform the next function evaluation(s). The potential of our approach is demonstrated using the so-called Sparse Pseudo-input GP as a global model. The algorithm is tested on four benchmark problems, whose number of starting points ranges from 102 to 104. Our results show that SpLEGO is effective and capable of solving problems with large number of starting points, and it even provides significant advantages when compared with state-of-the-art EI algorithms.

Sedimentological processes and environmental variability at Lake Ohrid (Macedonia, Albania) between 637 ka and the present

Lake Ohrid (Macedonia, Albania) is thought to be more than 1.2 million years old and host more than 300 endemic species. As a target of the International Continental scientific Drilling Program (ICDP), a successful deep drilling campaign was carried out within the scope of the Scientific Collaboration on Past Speciation Conditions in Lake Ohrid (SCOPSCO) project in 2013. Here, we present lithological, sedimentological, and (bio-)geochemical data from the upper 247.8 m composite depth of the overall 569 m long DEEP site sediment succession from the central part of the lake. According to an age model, which is based on 11 tephra layers (first-order tie points) and on tuning of bio-geochemical proxy data to orbital parameters (second-order tie points), the analyzed sediment sequence covers the last 637 kyr. The DEEP site sediment succession consists of hemipelagic sediments, which are interspersed by several tephra layers and infrequent, thin (< 5 cm) mass wasting deposits. The hemipelagic sediments can be classified into three different lithotypes. Lithotype 1 and 2 deposits comprise calcareous and slightly calcareous silty clay and are predominantly attributed to interglacial periods with high primary productivity in the lake during summer and reduced mixing during winter. The data suggest that high ion and nutrient concentrations in the lake water promoted calcite precipitation and diatom growth in the epilimnion during MIS15, 13, and 5. Following a strong primary productivity, highest interglacial temperatures can be reported for marine isotope stages (MIS) 11 and 5, whereas MIS15, 13, 9, and 7 were comparably cooler. Lithotype 3 deposits consist of clastic, silty clayey material and predominantly represent glacial periods with low primary productivity during summer and longer and intensified mixing during winter. The data imply that the most severe glacial conditions at Lake Ohrid persisted during MIS16, 12, 10, and 6, whereas somewhat warmer temperatures can be inferred for MIS14, 8, 4, and 2. Interglacial-like conditions occurred during parts of MIS14 and 8.

Crevassing and Calving of Glacial Ice

Calving of ice is a relatively new area of research in the still young field of glaciology. In the short time that calving has been studied, it has been mainly treated as an afterthought, with the predominant mode of thinking being that it will happen so to concern oneself with why is not important. Many studies dealt with observations of calving front positions over time vs. ice velocity in an attempt to quantify the calving rate as the difference between the two, while others have attempted to deduce some empirical relationship between calving rate and variables such as water depth or temperature. This study instead addresses the question of why, where, and when ice will first become crevassed, which is an obviously necessary condition for a later calving event to occur. Previous work examining the causes of calving used ideas put forth from a variety of fields, including civil engineering, materials science, and results from basic physics and mechanics. These theories are re-examined here and presented as part of a larger whole. Important results from the field of fracture mechanics are utilized frequently, and these results can be used as a predictor of ice behavior and intrinsic properties of ice, as well as properties like back stresses induced by local pinning points and resistive shears along glacial ice boundaries. A theory of fracture for a material experiencing creep is also presented with applications to ice shelves and crevasse penetration. Finally, a speculative theory regarding large scale iceberg formation is presented. It is meant mainly as an impetus to further discussion on the topic, with the hope that a model relating crevasse geometries to flow parameters can result in crevasse spacings that could produce the tabular icebergs which are so newsworthy. The primary focus of this thesis is to move away from the "after the fact" studies that are so common in calving research, and instead devote energy to determining what creates the conditions that drive the calving of ice in the first place.

Performance Evaluation of the New Connecticut Leading Employment Index Using Lead Profiles and BVAR Models

Dua and Miller (1996) created leading and coincident employment indexes for the state of Connecticut, following Moore's (1981) work at the national level. The performance of the Dua-Miller indexes following the recession of the early 1990s fell short of expectations. This paper performs two tasks. First, it describes the process of revising the Connecticut Coincident and Leading Employment Indexes. Second, it analyzes the statistical properties and performance of the new indexes by comparing the lead profiles of the new and old indexes as well as their out-of-sample forecasting performance, using the Bayesian Vector Autoregressive (BVAR) method. The new indexes show improved performance in dating employment cycle chronologies. The lead profile test demonstrates that superiority in a rigorous, non-parametric statistic fashion. The mixed evidence on the BVAR forecasting experiments illustrates the truth in the Granger and Newbold (1986) caution that leading indexes properly predict cycle turning points and do not necessarily provide accurate forecasts except at turning points, a view that our results support.

Global distributions of pteropods (Gymnosomata, Thecosomata, Pseudothecosomata) abundance and biomass - Gridded data product (NetCDF) - Contribution to the MAREDAT World Ocean Atlas of Plankton Functional Types

Pteropods are a group of holoplanktonic gastropods for which global biomass distribution patterns remain poorly resolved. The aim of this study was to collect and synthesize existing pteropod (Gymnosomata, Thecosomata and Pseudothecosomata) abundance and biomass data, in order to evaluate the global distribution of pteropod carbon biomass, with a particular emphasis on its seasonal, temporal and vertical patterns. We collected 25 902 data points from several online databases and a number of scientific articles. The biomass data has been gridded onto a 360 x 180° grid, with a vertical resolution of 33 WOA depth levels. Data has been converted to NetCDF format. Data were collected between 1951-2010, with sampling depths ranging from 0-1000 m. Pteropod biomass data was either extracted directly or derived through converting abundance to biomass with pteropod specific length to weight conversions. In the Northern Hemisphere (NH) the data were distributed evenly throughout the year, whereas sampling in the Southern Hemisphere was biased towards the austral summer months. 86% of all biomass values were located in the NH, most (42%) within the latitudinal band of 30-50° N. The range of global biomass values spanned over three orders of magnitude, with a mean and median biomass concentration of 8.2 mg C l-1 (SD = 61.4) and 0.25 mg C l-1, respectively for all data points, and with a mean of 9.1 mg C l-1 (SD = 64.8) and a median of 0.25 mg C l-1 for non-zero biomass values. The highest mean and median biomass concentrations were located in the NH between 40-50° S (mean biomass: 68.8 mg C l-1 (SD = 213.4) median biomass: 2.5 mg C l-1) while, in the SH, they were within the 70-80° S latitudinal band (mean: 10.5 mg C l-1 (SD = 38.8) and median: 0.2 mg C l-1). Biomass values were lowest in the equatorial regions. A broad range of biomass concentrations was observed at all depths, with the biomass peak located in the surface layer (0-25 m) and values generally decreasing with depth. However, biomass peaks were located at different depths in different ocean basins: 0-25 m depth in the N Atlantic, 50-100 m in the Pacific, 100-200 m in the Arctic, 200-500 m in the Brazilian region and >500 m in the Indo-Pacific region. Biomass in the NH was relatively invariant over the seasonal cycle, but more seasonally variable in the SH. The collected database provides a valuable tool for modellers for the study of ecosystem processes and global biogeochemical cycles.

Effect of Prosopis flexuosa on understory species and its importance to pastoral management in woodlands of the Central Monte Desert

In the Monte Biogeographic Province, located in the arid region of Argentina, the presence of Prosopis flexuosa DC. produces spatial heterogeneity through edaphic modifications and microclimate changes. This results in vegetation patches differing in species composition and abundance. However, this interaction can be modified by the occurrence of gradients of biotic stress or disturbance intensity. In particular, grazing has been observed to enhance or reduce vegetation heterogeneity. Such complex of interactions could determine forage availability for cattle in one of the driest areas of the Monte Desert. We assessed the effect of Prosopis on understory species and analyzed whether the outcomes of this interaction differed with distance to watering points, as a proxy of grazing intensity, in the Northeast of Mendoza Province, Argentina. We used a two-way factorial design including the following factors: 1) microsite (under the cover of P. flexuosa trees and in intercanopy microsites) and 2) distance to watering points ("near the watering point", 500-700 m away, and "far from the watering point", 3-4 km away). Cover of each species, total cover, bare soil, and litter were recorded, and plant diversity, richness, and evenness were estimated with the modified Point Quadrat method. Results showed that P. flexuosa cover, distance from watering points, and the interaction between them determined species composition, abundance and spatial distribution of understory species, and were, consequently, a determining factor for forage availability. The presence of P. flexuosa enhances carrying capacity by supporting higher abundance of grasses under its canopy. Near watering points, high grazing intensity appears to disrupt the patches formed under P. flexuosa canopies, reducing the differences between microsites.

Calcareous nannofossil stratigraphy, mass accumulation rates and bulk density of sediments from ODP Leg 173 sites

Drilling on the Iberia Abyssal Plain during Ocean Drilling Program Leg 173 allowed us to recover Upper Cretaceous through Paleocene sediments at Sites 1068 and 1069 and only upper Paleocene sediments at Site 1067, which expands considerably the Upper Cretaceous to Paleocene record for this region. Of these three sites, Site 1068 recovered uppermost Cretaceous sediments as well as the most complete Paleocene record, whereas Site 1067 yielded only uppermost Paleocene sediments (Zone CP8). Site 1069 provided a rather complete upper Campanian through Maastrichtian section but a discontinuous Paleocene record. After a detailed calcareous nannofossil biostratigraphy was documented in distribution charts, we calculated mass accumulation rates for Holes 1068A and 1069A. Sediments in Hole 1068A apparently record the final stages of burial of a high basement block by turbidity flows. Accumulation rates through the Upper Cretaceous indicate relatively high rates, 0.95 g/cm**2/k.y., but may be unreliable because of the lack of datum points and/or possible hiatuses. Accumulation rates in the Paleocene section of Hole 1068A fluctuated every few million years from lower (~0.35 g/cm**2/k.y.) to higher rates (~0.85 g/cm**2/k.y.) until the latest Paleocene, when rates increased to an average of ~2.0 g/cm**2/k.y. Mass accumulation rates for the Upper Cretaceous in Hole 1069A indicate a steady rate of ~0.60 g/cm**2/k.y. from 75 to 72 Ma. There may have been one or more hiatuses between 72 and 68 Ma (combined Zone CC24 through Subzone CC25b), as indicated by the very low accumulation rate of 0.15 g/cm**2/k.y. The Paleocene section of Hole 1069A does not show the same continuous record, which may result from fluctuations in the carbonate compensation depth and poor recovery (average = 40%). Zones CP4 and CP5 are missing within a barren interval; this and numerous other barren intervals affect the precision of the nannofossil zonation and calculation of mass accumulation rates. However, in spite of these missing zones, mass accumulation rates do not seem to indicate the presence of hiatuses as the rates for this barren interval average ~1.0 g/cm**2/k.y. This study set out to test the hypothesis that a reliable biostratigraphic record could be constructed from sediments derived from turbidity flows deposited below the carbonate compensation depth. As illustrated here, not only could a reliable biostratigraphic record be determined from these sediments, but sedimentation and mass accumulation rates could also be determined, allowing inferences to be drawn concerning the sedimentary history of this passive margin. The reliability of this record is confirmed by independent verification by the establishment of a magnetostratigraphy for the same cores.

How do free trade agreements reduce tariff rates and non-tariff barriers?

This paper empirically investigates how far free trade agreements (FTAs) successfully lower tariff rates and non-tariff barriers (NTBs) for manufacturing industries by employing the bilateral tariff and NTB data in a time series for countries around the world. We find that FTAs under GATT Article XXIV and the Enabling Clause contribute to reducing tariff rates by 2.1% points and 1.5% points, respectively. In the case of NTBs, their respective impacts are 6.6% points and 5.7% points. Membership in the World Trade Organization (WTO) does not contribute greatly to reducing tariff rates but does play a significant role in reducing NTBs. These results provide important implications for the literature on numerical assessments of FTAs.

Integrating software testing and run-time checking in an assertion verification framework

We have designed and implemented a framework that unifies unit testing and run-time verification (as well as static verification and static debugging). A key contribution of our approach is that a unified assertion language is used for all of these tasks. We first propose methods for compiling runtime checks for (parts of) assertions which cannot be verified at compile-time via program transformation. This transformation allows checking preconditions and postconditions, including conditional postconditions, properties at arbitrary program points, and certain computational properties. The implemented transformation includes several optimizations to reduce run-time overhead. We also propose a minimal addition to the assertion language which allows defining unit tests to be run in order to detect possible violations of the (partial) specifications expressed by the assertions. This language can express for example the input data for performing the unit tests or the number of times that the unit tests should be repeated. We have implemented the framework within the Ciao/CiaoPP system and effectively applied it to the verification of ISO-prolog compliance and to the detection of different types of bugs in the Ciao system source code. Several experimental results are presented that ¡Ilústrate different trade-offs among program size, running time, or levéis of verbosity of the messages shown to the user.

Integrating software testing and run-time checking in an assertion verification framework.

We have designed and implemented a framework that unifies unit testing and run-time verification (as well as static verification and static debugging). A key contribution of our approach is that a unified assertion language is used for all of these tasks. We first propose methods for compiling runtime checks for (parts of) assertions which cannot be verified at compile-time via program transformation. This transformation allows checking preconditions and postconditions, including conditional postconditions, properties at arbitrary program points, and certain computational properties. The implemented transformation includes several optimizations to reduce run-time overhead. We also propose a minimal addition to the assertion language which allows defining unit tests to be run in order to detect possible violations of the (partial) specifications expressed by the assertions. This language can express for example the input data for performing the unit tests or the number of times that the unit tests should be repeated. We have implemented the framework within the Ciao/CiaoPP system and effectively applied it to the verification of ISO-prolog compliance and to the detection of different types of bugs in the Ciao system source code. Several experimental results are presented that illustrate different trade-offs among program size, running time, or levels of verbosity of the messages shown to the user.

Instability and topology bifurcations on a hemisphere-cylinder at high angle of attack

La configuración de un cilindro acoplado a una semi-esfera, conocida como ’hemispherecylinder’, se considera como un modelo simplificado para numerosas aplicaciones industriales tales como fuselaje de aviones o submarinos. Por tanto, el estudio y entendimiento de los fenómenos fluidos que ocurren alrededor de dicha geometría presenta gran interés. En esta tesis se muestra la investigación del origen y evolución de los, ya conocidos, patrones de flujo (burbuja de separación, vórtices ’horn’ y vórtices ’leeward’) que se dan en esta geometría bajo condiciones de flujo separado. Para ello se han llevado a cabo simulaciones numéricas (DNS) y ensayos experimentales usando la técnica de Particle Image Velocimetry (PIV), para una variedad de números de Reynolds (Re) y ángulos de ataque (AoA). Se ha aplicado sobre los resultados numéricos la teoría de puntos críticos obteniendo, por primera vez para esta geometría, un diagrama de bifurcaciones que clasifica los diferentes regímenes topológicos en función del número de Reynolds y del ángulo de ataque. Se ha llevado a cabo una caracterización completa sobre el origen y la evolución de los patrones estructurales característicos del cuerpo estudiado. Puntos críticos de superficie y líneas de corriente tridimensionales han ayudado a describir el origen y la evolución de las principales estructuras presentes en el flujo hasta alcanzar un estado de estabilidad desde el punto de vista topológico. Este estado se asocia con el patrón de los vórtices ’horn’, definido por una topología característica que se encuentra en un rango de números de Reynolds muy amplio y en regímenes compresibles e incompresibles. Por otro lado, con el objeto de determinar las estructuras presentes en el flujo y sus frecuencias asociadas, se han usado distintas técnicas de análisis: Proper Orthogonal Decomposition (POD), Dynamic Mode Decomposition (DMD) y análisis de Fourier. Dichas técnicas se han aplicado sobre los datos experimentales y numéricos, demostrándose la buena concordancia entre ambos resultados. Finalmente, se ha encontrado en ambos casos, una frecuencia dominante asociada con una inestabilidad de los vórtices ’leeward’. ABSTRACT The hemisphere-cylinder may be considered as a simplified model for several geometries found in industrial applications such as aircrafts’ fuselages or submarines. Understanding the complex flow phenomena that surrounds this particular geometry is therefore of major industrial interest. This thesis presents an investigation of the origin and evolution of the complex flow pattern; i.e. separation bubbles, horn vortices and leeward vortices, around the hemisphere-cylinder under separated flow conditions. To this aim, threedimensional Direct Numerical Simulations (DNS) and experimental tests, using Particle Image Velocimetry (PIV) techniques, have been performed for a variety of Reynolds numbers (Re) and angles of attack (AoA). Critical point theory has been applied to the numerical simulations to provide, for the first time for this geometry, a bifurcation diagram that classifies the different flow topology regimes as a function of the Reynolds number and the angle of attack. A complete characterization about the origin and evolution of the complex structural patterns of this geometry has been put in evidence. Surface critical points and surface and volume streamlines were able to describe the main flow structures and their strong dependence with the flow conditions up to reach the structurally stable state. This state was associated with the pattern of the horn vortices, found on ranges from low to high Reynolds numbers and from incompressible to compressible regimes. In addition, different structural analysis techniques have been employed: Proper Orthogonal Decomposition (POD), Dynamic Mode Decomposition (DMD) and Fourier analysis. These techniques have been applied to the experimental and numerical data to extract flow structure information (i.e. modes and frequencies). Experimental and numerical modes are shown to be in good agreement. A dominant frequency associated with an instability of the leeward vortices has been identified in both, experimental and numerical results.

Middleware architectures for the Smart Grid : survey and challenges in the foreseeable future

The traditional power grid is just a one-way supplier that gets no feedback data about the energy delivered, what tariffs could be the most suitable ones for customers, the shifting daily needs of electricity in a facility, etc. Therefore, it is only natural that efforts are being invested in improving power grid behavior and turning it into a Smart Grid. However, to this end, several components have to be either upgraded or created from scratch. Among the new components required, middleware appears as a critical one, for it will abstract all the diversity of the used devices for power transmission (smart meters, embedded systems, etc.) and will provide the application layer with a homogeneous interface involving power production and consumption management data that were not able to be provided before. Additionally, middleware is expected to guarantee that updates to the current metering infrastructure (changes in service or hardware availability) or any added legacy measuring appliance will get acknowledged for any future request. Finally, semantic features are of major importance to tackle scalability and interoperability issues. A survey on the most prominent middleware architectures for Smart Grids is presented in this paper, along with an evaluation of their features and their strong points and weaknesses.

Parametric Generation of Planing Hulls with NURBS Surfaces

This article presents a mathematical method for producing hard-chine ship hulls based on a set of numerical parameters that are directly related to the geometric features of the hull and uniquely define a hull form for this type of ship. The term planing hull is used generically to describe the majority of hard-chine boats being built today. This article is focused on unstepped, single-chine hulls. B-spline curves and surfaces were combined with constraints on the significant ship curves to produce the final hull design. The hard-chine hull geometry was modeled by decomposing the surface geometry into boundary curves, which were defined by design constraints or parameters. In planing hull design, these control curves are the center, chine, and sheer lines as well as their geometric features including position, slope, and, in the case of the chine, enclosed area and centroid. These geometric parameters have physical, hydrodynamic, and stability implications from the design point of view. The proposed method uses two-dimensional orthogonal projections of the control curves and then produces three-dimensional (3-D) definitions using B-spline fitting of the 3-D data points. The fitting considers maximum deviation from the curve to the data points and is based on an original selection of the parameterization. A net of B-spline curves (stations) is then created to match the previously defined 3-D boundaries. A final set of lofting surfaces of the previous B-spline curves produces the hull surface.

Traffic sign detection and tracking using robust 3D analysis

In this paper we present an innovative technique to tackle the problem of automatic road sign detection and tracking using an on-board stereo camera. It involves a continuous 3D analysis of the road sign during the whole tracking process. Firstly, a color and appearance based model is applied to generate road sign candidates in both stereo images. A sparse disparity map between the left and right images is then created for each candidate by using contour-based and SURF-based matching in the far and short range, respectively. Once the map has been computed, the correspondences are back-projected to generate a cloud of 3D points, and the best-fit plane is computed through RANSAC, ensuring robustness to outliers. Temporal consistency is enforced by means of a Kalman filter, which exploits the intrinsic smoothness of the 3D camera motion in traffic environments. Additionally, the estimation of the plane allows to correct deformations due to perspective, thus easing further sign classification.

Effects of coaches' timeouts on basketball teams' offensive and defensive performances according to momentary differences in score and game period.

The aim of the present study was to assess the effects of game timeouts on basketball teams? offensive and defensive performances according to momentary differences in score and game period. The sample consisted of 144 timeouts registered during 18 basketball games randomly selected from the 2007 European Basketball Championship (Spain). For each timeout, five ball possessions were registered before (n?493) and after the timeout (n?475). The offensive and defensive efficiencies were registered across the first 35 min and last 5 min of games. A k-means cluster analysis classified the timeouts according to momentary score status as follows: losing ( ?10 to ?3 points), balanced ( ?2 to 3 points), and winning (4 to 10 points). Repeated-measures analysis of variance identified statistically significant main effects between pre and post timeout offensive and defensive values. Chi-square analysis of game period identified a higher percentage of timeouts called during the last 5 min of a game compared with the first 35 min (64.999.1% vs. 35.1910.3%; x ?5.4, PB0.05). Results showed higher post timeout offensive and defensive performances. No other effect or interaction was found for defensive performances. Offensive performances were better in the last 5 min of games, with the least differences when in balanced situations and greater differences when in winning situations. Results also showed one interaction between timeouts and momentary differences in score, with increased values when in losing and balanced situations but decreased values when in winning situations. Overall, the results suggest that coaches should examine offensive and defensive performances according to game period and differences in score when considering whether to call a timeout.